Inhomogeneity-induced magnetoresistance (IMR) reported in some non-magnetic semiconductors1, 2, 3, 4, 5, 6, 7, 8, particularly silicon1, 6, 7, 8, has generated considerable interest owing to the large magnitude of the effect and its linear field dependence (albeit at high magnetic fields). Various theories implicate9, 10, 11, 12, 13, 14, 15, 16, 17, 18 spatial variation of the carrier mobility as being responsible for IMR. Here we show that IMR in lightly doped silicon can be significantly enhanced through hole injection, and then tuned by an applied current to arise at low magnetic fields. In our devices, the ‘inhomogeneity’ is provided by the p–n boundary formed between regions where conduction is dominated by the minority and majority charge carriers (holes and electrons) respectively; application of a magnetic field distorts the current in the boundary region, resulting in large magnetoresistance. Because this is an intrinsically spatial effect, the geometry of the device can be used to enhance IMR further: we designed an IMR device whose room-temperature field sensitivity at low fields was greatly improved, with magnetoresistance reaching 10% at 0.07 T and 100% at 0.2 T, approaching the performance of commercial giant-magnetoresistance devices19, 20. The combination of high sensitivity to low magnetic fields and large high-field response should make this device concept attractive to the magnetic-field sensing industry. Moreover, because our device is based on a conventional silicon platform, it should be possible to integrate it with existing silicon devices and so aid the development of silicon-based magnetoelectronics.
Figures at a glance
Figure 1: I–V characteristics and Hall coefficient measured in In/SiO2/Si/SiO2/In at 300 K. 
a, Measurement geometry. The width W is 3.0 mm, the distance between the voltage electrodes L is 2.3 mm and the lateral distance between the current injecting electrode and the Hall electrodes D is 3.2 mm. b, The closed and open circles show the I–V characteristics and Hall coefficient of sample 15, respectively. Insets show the locations of the p–n boundary.
Figure 2: I–V characteristics under magnetic field and the magnetoresistance of sample 20 at 300 K. 
a, I–V characteristics of sample 20 under different magnetic fields. The current range of the transition region shifted towards a lower-current region as the magnetic field increased. The inset shows the geometry of the sample. b, The magnetic field dependence of magnetoresistance in sample 20. Between 146 μA and 226 μA, a magnetoresistance transition occurred from normal to abnormal. Above the transition, abnormal magnetoresistance was also observed.
Figure 3: Potential (colour scale) and current density (arrows) distributions and the effect of electrode geometry on magnetoresistance. 
The solid vertical lines in a and b show the p–n boundary. a, B = 0. The potential was symmetric about the x axis and current flowed along the x axis. b, B = μ−1. A positive Hall voltage appeared in the p region and a negative Hall voltage appeared in the n region. The boundary acted as a magnetic scattering resource. The current trajectory was distorted upward. c, As y0 increased, the apparent maximum magnetoresistance of sample 18 increased. d, As x0 decreased, the apparent maximum magnetoresistance of sample 26 increased. Insets in c and d show the placement of voltage contacts.
Figure 4: Magnetoresistance of sample 40 at 300 K. 
ftp://mail.ihim.uran.ru/localfiles/nature10375.pdf
